yihua commented on code in PR #17725:
URL: https://github.com/apache/hudi/pull/17725#discussion_r2795557113
##########
hudi-utilities/src/main/java/org/apache/hudi/utilities/sources/debezium/DebeziumSource.java:
##########
@@ -124,7 +122,7 @@ protected Pair<Option<Dataset<Row>>, Checkpoint>
fetchNextBatch(Option<Checkpoin
LOG.info("About to read " + totalNewMsgs + " from Kafka for topic :" +
offsetGen.getTopicName());
try {
- String schemaStr =
schemaRegistryProvider.fetchSchemaFromRegistry(getStringWithAltKeys(props,
HoodieSchemaProviderConfig.SRC_SCHEMA_REGISTRY_URL));
+ String schemaStr =
schemaProvider.getSourceHoodieSchema().getAvroSchema().toString();
Review Comment:
`getSourceHoodieSchema()` can return null (when the underlying
`getSourceSchema()` returns null), which would cause an NPE here. Could you add
a null check, or at least call it once and verify it's non-null before chaining
`.getAvroSchema().toString()`?
##########
hudi-utilities/src/main/java/org/apache/hudi/utilities/sources/debezium/DebeziumSource.java:
##########
@@ -100,15 +99,14 @@ public DebeziumSource(TypedProperties props,
JavaSparkContext sparkContext,
throw new HoodieReadFromSourceException(error, e);
}
- // Currently, debezium source requires Confluent/Kafka schema-registry to
fetch the latest schema.
- if (schemaProvider == null || !(schemaProvider instanceof
SchemaRegistryProvider)) {
- schemaRegistryProvider = new SchemaRegistryProvider(props, sparkContext);
+ if (schemaProvider == null) {
+ this.schemaProvider = new SchemaRegistryProvider(props, sparkContext);
} else {
- schemaRegistryProvider = (SchemaRegistryProvider) schemaProvider;
+ this.schemaProvider = schemaProvider;
Review Comment:
Add a test case around this change?
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
